A research article published in Volume 17, Issue 11 of Aging-US on 25 November 2025 presents a large-scale, AI-based analysis of how ageing research has evolved over the past century. Titled “A natural language processing–driven map of the aging research landscape, the paper adopts a meta-research perspective, using computational tools to examine not a single biological problem, but the structure, priorities, and blind spots of ageing science as a whole. It aims to provide a clear, data-driven overview of where the field has come from and where it may need to go next.
The study was led by Jose Perez-Maletzki, affiliated with Universidad Europea de Valencia and Universitat de València, in collaboration with Jorge Sanz-Ros from Stanford University School of Medicine. Together, the authors used artificial intelligence to analyse global ageing research output, seeking to identify long-term trends, shifts in emphasis, and areas that remain underexplored. Their approach reflects a growing recognition that understanding the organisation of scientific knowledge is itself essential for accelerating progress.
To conduct the analysis, the team examined more than 460,000 scientific abstracts published between 1925 and 2023. This massive dataset allowed them to capture nearly a century of research across basic biology, clinical medicine, and population health. By focusing on abstracts, the authors were able to include a broad range of disciplines and journals, producing a more comprehensive and less selective overview than is typically possible with traditional narrative reviews.
Using natural language processing and machine-learning techniques, the researchers grouped publications into thematic clusters and tracked how interest in each topic changed over time. Their methodology combined Latent Dirichlet Allocation, term frequency–inverse document frequency analysis, dimensionality reduction, and clustering to generate a structured thematic map of ageing research. This enabled them to visualise both dominant research areas and weaker connections between topics, revealing patterns that are difficult to detect through manual review alone.
One key finding was a clear historical shift in focus. Early ageing research was primarily centred on fundamental cellular processes and animal models, whereas more recent decades show a strong move towards clinical and disease-oriented studies. In particular, research on Alzheimer’s disease, dementia, and geriatric healthcare has grown rapidly. While this reflects the realities of ageing populations, the analysis also indicates a widening gap between basic biological research and clinical application.
The study shows that basic and clinical ageing research often develops in parallel, with limited integration. Clinical studies tend to focus on care, neurodegeneration, and age-related disease, while basic science emphasises mechanisms such as mitochondrial dysfunction, telomere shortening, oxidative stress, and cellular senescence. Emerging areas like autophagy, RNA biology, and nutrient sensing are expanding quickly but remain weakly connected to clinical research. The authors argue that these gaps represent missed opportunities and potential directions for future interdisciplinary work.
Overall, this AI-driven analysis provides a powerful tool for reflecting on how ageing research is organised and prioritised. By identifying both strong connections and neglected links, the study offers guidance for shaping more integrated and translational research strategies. As global populations continue to age, such a comprehensive and critical overview may help ensure that future research is not only productive but also better aligned with real-world health outcomes.
More information: Jose Perez-Maletzki et al, A natural language processing–driven map of the aging research landscape, Aging-US. DOI: 10.18632/aging.206340
Journal information: Aging-US Provided by Impact Journals LLC
